How we rank AI courses

What we track

We track courses on the platforms where learners leave public ratings: Udemy, Coursera and DataCamp. We also track notable free courses from DeepLearning.AI, Anthropic, Hugging Face, Google and fast.ai. A course enters a list when it teaches that topic directly, is in English and, on Udemy, has at least 300 ratings. We remove courses that are retired or no longer match the topic.

AETW earns a commission when you buy through links to Udemy, Coursera and DataCamp, and most tracked courses are on those platforms. We decide which courses to track. We do not decide their order: the score does.

The score

Every course gets a score out of 100, recalculated every week:

• Learner rating, 30%. The platform's average rating, adjusted toward 4.4 stars when a course has few ratings, so a 4.9 from 40 people doesn't beat a 4.7 from 50,000. 3.8 stars scores 0 and 4.9 scores 100.
• Popularity, 20%. The number of ratings on a log scale: 10 ratings scores 0 and 100,000 scores 100. Where a platform shows no ratings, we use enrollment.
• Momentum, 20%. New ratings since last week's check on a log scale, so a course that is actively being taken now scores higher. 1,000 new ratings in a week scores 100. A course's first week scores a neutral 50.
• Freshness, 15%. Courses updated in the last two months score 100, falling to 0 after two years without an update. AI tools change quickly, so an old course loses points even if its rating is high.
• Curriculum, 15%. Our 0–10 review of the published syllabus: depth, hands-on projects and how current the tools are. This is the only editorial input, and we show the verdict on every entry.

Courses with no public rating data score a neutral 50 on rating, popularity and momentum, so they are ranked mostly on curriculum and freshness. Ties go to the course with more ratings.

What we don't do

• Commission rates and affiliate status are not inputs to the score.
• Nobody can pay for a place on these lists.
• We don't claim to have taken every course. Verdicts are based on the syllabus and course materials.

Where the data comes from

Ratings, rating counts, enrollment and update dates are read from each course's public page every week. Ratings belong to the platforms and their learners. If a number looks wrong, tell us and we'll check it before the next update.